Geographic analysis of immunization patterns in Genesee County

Kyle S. Enger, MPH

Vaccine-Preventable Disease Epidemiologist

Division of Immunization

Michigan Dept. of Community Health (MDCH)

Data source: MCIR(Michigan Care Improvement Registry)

• Operational since 1998

• Birth-to-death immunization registry– < 20 y age restriction removed June 2006

• ~95% of providers participate

• Part of CDC’s IIS sentinel site project

• Michigan’s population is large– 10 million people– 130,000 births/year

Methods: Geocoding

• MCIR receives ‘responsible party’ addresses from birth certificate or clinicians

• Address is converted to geographic coordinates by the MI Dept. of Human Services (DHS) using Postalsoft

• ~85% of addresses yield good coordinates

• Very easy to view using GIS (ArcView)

Methods: Density mapping

• Converts point data to a continuous surface

• Similar to a drip castle – each point gets a drip, and drips can stack on each other

• ‘High’ areas are bright; ‘low’ areas are dark

• Brightness=kids/mile2

• ArcGIS Spatial Analyst

Methods: Cluster analysis

• Determines where a factor is more common than population would suggest

• Generates circular clusters; provides relative risk & P-value for each cluster

• SaTScan (free software)

Study population

• Children with a Genesee County address & geographic coordinates: – aged 19-35 months on 28 Feb. 2007– on record in MCIR

• Density maps and cluster analysis:– For various childhood vaccines

• DTaP, MMR, polio, Hib, hep. B, varicella, PCV7

• Completely unimmunized

• Partially immunized

Population density

Density of kids with 0 DTaP - Genesee

Density of kids with 0 DTaP - Flint

Density of DTaP needed - Genesee

Density of DTaP needed - Flint

DTaP clusteringFor illustration only. Cluster locations & sizes have been randomly changed.

PCV7 clusteringFor illustration only. Cluster locations & sizes have been randomly changed.

Questions?

Kyle Enger

engerk@michigan.gov

Acknowledgements

Mark Valacak, Sherry Wood, & Fatema Mamou, Genesee County Health Dept.

Kevin Czubachowski, Mich. Dept. Comm. Health

SaTScan settings

• Max cluster size: 50% of population

• Clusters of high RR only

• Bernoulli model (1/0)– (immunized/unimmunized)

• No pairs of centers both in each other’s clusters

Density map settings

• Density type: kernel

• Cell size: 50m

• Kernel radius: 1 km

• Output area units: kids/mile2 or iz./mile2

• Color steps: Factor of 1.67